From daria at cs.cmu.edu Wed Nov 4 09:25:40 2009 From: daria at cs.cmu.edu (Daria Sorokina) Date: Wed, 4 Nov 2009 09:25:40 -0500 Subject: [Research] Auton Lab meeting next week In-Reply-To: <4AEAFB65.1000502@cs.cmu.edu> References: <4AEAFB65.1000502@cs.cmu.edu> Message-ID: <00d101ca5d5a$af9dbab0$0ed93010$@cmu.edu> Reminder - the talk is today. Daria -----Original Message----- From: research-bounces at autonlab.org [mailto:research-bounces at autonlab.org] On Behalf Of Artur Dubrawski Sent: Friday, October 30, 2009 10:43 AM To: research at autonlab.org Subject: [Research] Auton Lab meeting next week Dear Autonians, Now with the fall foliage past its peak, everyone in Pittsburgh can see more structural detail of trees growing around. It feels appropriate and timely for computer scientists to relate to this phenomenon. Fundamental questions, such as why the "real" trees have roots in the bottom and leaves up on top, entirely opposite to how they are in computers, as well as other crucial and relevant issues, will be addressed by the leading Auton Lab tree grower Dr. Daria Sorokina. When: Wednesday November 4th at noon Where: NSH 1507 Food: yes Abstract: see below See you all there! Artur ---- TreeExtra package: prediction and descriptive analysis with ensembles of trees. In this talk I am going to describe a C++ package I have been developing over the past two years. TreeExtra contains a number of general data mining tools relevant for a large range of problems. - Additive Groves A powerful ensemble of trees, significantly outperforms other ensembles on regression and is highly competitive on classification problems. TreeExtra implementation provides several modes of training, automatic parameter choice and parallelization tools. - Bagged trees with fast feature evaluation. Bagged trees is a fast and simple ensemble, in our implementation it is equipped with built-in feature importance ranking. - Backward elimination feature selection Thorough feature selection, leaves only a small set of truly important non-redundant features. - Interaction detection Detects which features are involved in complex effects and need to be studied together - Effect plots (or partial dependence plots) Visualization of an effect of a single variable in a given model - Interaction plots Visualization of a joint effect of two variables in a model along with the joint distribution of those variables _______________________________________________ Research mailing list Research at autonlab.org https://www.autonlab.org/mailman/listinfo/research From awd at cs.cmu.edu Mon Nov 9 11:11:37 2009 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Mon, 09 Nov 2009 11:11:37 -0500 Subject: [Research] Auton Lab meeting: this Wednesday (Nov 11th, 12noon, Gates 6501) Message-ID: <4AF83F39.2050804@cs.cmu.edu> Dear Autonians, This week we will hear Purna Sarkar talk about her current work into useful and manageable analysis or large networks. The title and abstract will be provided soon by the presenter herself. Place: Gates 6501 Time: noon, Nov 11th Food: yes See you all there Artur From krw at andrew.cmu.edu Mon Nov 9 11:41:14 2009 From: krw at andrew.cmu.edu (Karen Widmaier) Date: Mon, 9 Nov 2009 11:41:14 -0500 Subject: [Research] Auton Lab meeting: this Wednesday (Nov 11th, 12noon, Gates 6501) In-Reply-To: <4AF83F39.2050804@cs.cmu.edu> References: <4AF83F39.2050804@cs.cmu.edu> Message-ID: <016b01ca615b$73d15fa0$5b741ee0$@cmu.edu> I was able to make a switch with someone, so we will be in NSH 1507 on Wednesday, November 11. Karen -----Original Message----- From: research-bounces at autonlab.org [mailto:research-bounces at autonlab.org] On Behalf Of Artur Dubrawski Sent: Monday, November 09, 2009 11:12 AM To: research at autonlab.org Subject: [Research] Auton Lab meeting: this Wednesday (Nov 11th, 12noon, Gates 6501) Dear Autonians, This week we will hear Purna Sarkar talk about her current work into useful and manageable analysis or large networks. The title and abstract will be provided soon by the presenter herself. Place: Gates 6501 Time: noon, Nov 11th Food: yes See you all there Artur _______________________________________________ Research mailing list Research at autonlab.org https://www.autonlab.org/mailman/listinfo/research From awd at cs.cmu.edu Tue Nov 10 13:02:13 2009 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Tue, 10 Nov 2009 13:02:13 -0500 Subject: [Research] RESCHEDULED Re: Auton Lab meeting: this Wednesday (Nov 11th, 12noon, Gates 6501) In-Reply-To: <4AF83F39.2050804@cs.cmu.edu> References: <4AF83F39.2050804@cs.cmu.edu> Message-ID: <4AF9AAA5.2080803@cs.cmu.edu> Hello, The meeting tomorrow is CANCELED b/c Purna is down with cold (get well soon Purna!). We have tentatively rescheduled it for Monday, Nov 16th, lunchtime. Details will be provided when available. Artur Artur Dubrawski wrote: > Dear Autonians, > > This week we will hear Purna Sarkar talk about her current > work into useful and manageable analysis or large networks. > The title and abstract will be provided soon by the presenter > herself. > > Place: Gates 6501 > Time: noon, Nov 11th > Food: yes > > See you all there > Artur > > > _______________________________________________ > Research mailing list > Research at autonlab.org > https://www.autonlab.org/mailman/listinfo/research > > From awd at cs.cmu.edu Sun Nov 15 09:54:18 2009 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Sun, 15 Nov 2009 09:54:18 -0500 Subject: [Research] CONFIRMED : lab meeting on *Monday* In-Reply-To: <4AF9AAA5.2080803@cs.cmu.edu> References: <4AF83F39.2050804@cs.cmu.edu> <4AF9AAA5.2080803@cs.cmu.edu> Message-ID: <4B00161A.7000505@cs.cmu.edu> Purna's talk is going to happen! Where: NSH 1507 When: MONDAY 11/16 at noon Food: yes See you all there Artur Artur Dubrawski wrote: > Hello, > > The meeting tomorrow is CANCELED b/c Purna is down with cold > (get well soon Purna!). > > We have tentatively rescheduled it for Monday, Nov 16th, lunchtime. > Details will be provided when available. > > Artur > > Artur Dubrawski wrote: >> Dear Autonians, >> >> This week we will hear Purna Sarkar talk about her current >> work into useful and manageable analysis or large networks. >> The title and abstract will be provided soon by the presenter >> herself. >> >> Place: Gates 6501 >> Time: noon, Nov 11th >> Food: yes >> >> See you all there >> Artur >> >> >> _______________________________________________ >> Research mailing list >> Research at autonlab.org >> https://www.autonlab.org/mailman/listinfo/research >> >> > > From psarkar at cs.cmu.edu Sun Nov 15 22:34:18 2009 From: psarkar at cs.cmu.edu (Purnamrita Sarkar) Date: Sun, 15 Nov 2009 22:34:18 -0500 (EST) Subject: [Research] CONFIRMED : lab meeting on *Monday* In-Reply-To: <4B00161A.7000505@cs.cmu.edu> References: <4AF83F39.2050804@cs.cmu.edu> <4AF9AAA5.2080803@cs.cmu.edu> <4B00161A.7000505@cs.cmu.edu> Message-ID: <61106.71.240.67.99.1258342458.squirrel@webmail.cs.cmu.edu> Hi All, Here are the title and abstract: A Scalable External Memory Graph Partitioning Algorithm to Reduce Diskseeks in Random Walks This paper introduces an algorithm to generate efficient disk-resident representations of graphs that are too large to fit in main memory. The representation is efficient in the sense that stepping through paths of connected nodes in the graph (e.g. simulating a random walk) involves relatively few disk seeks. Since a broad class of graph-based similarity metrics (e.g. personalized pagerank, hitting and commute times, simrank) can be estimated by simulating random walks in a graph, this algorithm will strongly benefi t applications such as link prediction in social networks, personalized graph search, fraud detection and collaborative ltering. We build on previous work that has shown that personalized pagerank based graph-partitioning has desirable properties, and our primary contribution is an efficient algorithm based purely on sequential sweeps through les to generate such partitions. In addition to the disk-based algorithm, this paper provides error bounds for computing approximate personalized pagerank vectors, which can be used to find good local partitions (graph cuts with small conductance) near a given starting point in a graph. We also present some new theoretical properties of personalized pagerank which enables us to scale our algorithm up-to real world graphs with up-to 69 million edges. We empirically demonstrate the utility of the disk-based partitioning for link prediction tasks on real world disk-resident graphs like DBLP, Citeseer and Live- Journal. In addition to vanilla random walk simulations, we present deterministic algorithms for quickly computing top k neighbors of a node in personalized pagerank using our disk-resident graph representation. Purna On Sun, November 15, 2009 9:54 am, Artur Dubrawski wrote: > Purna's talk is going to happen! > > > Where: NSH 1507 > When: MONDAY 11/16 at noon > Food: yes > > > See you all there > Artur > > > > Artur Dubrawski wrote: > >> Hello, >> >> >> The meeting tomorrow is CANCELED b/c Purna is down with cold >> (get well soon Purna!). >> >> >> We have tentatively rescheduled it for Monday, Nov 16th, lunchtime. >> Details will be provided when available. >> >> >> Artur >> >> >> Artur Dubrawski wrote: >> >>> Dear Autonians, >>> >>> >>> This week we will hear Purna Sarkar talk about her current >>> work into useful and manageable analysis or large networks. The title >>> and abstract will be provided soon by the presenter herself. >>> >>> Place: Gates 6501 >>> Time: noon, Nov 11th >>> Food: yes >>> >>> >>> See you all there >>> Artur >>> >>> >>> >>> _______________________________________________ >>> Research mailing list >>> Research at autonlab.org >>> https://www.autonlab.org/mailman/listinfo/research >>> >>> >>> >> >> > > _______________________________________________ > Research mailing list > Research at autonlab.org > https://www.autonlab.org/mailman/listinfo/research > > > From awd at cs.cmu.edu Mon Nov 16 02:38:42 2009 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Mon, 16 Nov 2009 02:38:42 -0500 Subject: [Research] one more Lab meeting this week Message-ID: <4B010182.5040208@cs.cmu.edu> Dear Autonians, On Wednesday, we will be in for a rare treat. Three tenors, Skyler, Ed, and Daniel, will be performing a piece in three movements on the topic of "Fast Subset Scanning and Summing for Scalable Pattern Detection". This will be a dress rehearsal for the upcoming tournee in Florida. Please come in big numbers, and enjoy. Time: Noon on Wednesday, Nov 18. Karen W. will follow up with the location info. Artur From siddiqi at cs.cmu.edu Mon Nov 16 12:10:37 2009 From: siddiqi at cs.cmu.edu (Sajid Siddiqi) Date: Mon, 16 Nov 2009 12:10:37 -0500 (EST) Subject: [Research] Thin SVD code in auton codebase? Message-ID: <3875.128.2.182.68.1258391437.squirrel@webmail.cs.cmu.edu> Hi, Does anyone know if there is code for doing a thin SVD (i.e. top k singular values and vectors) in the auton codebase? The svd.c code in /utils seems to return the full SVD component matrices for an m x n matrix, i.e. a m x n matrix U, a n x n diagonal matrix S (as a dyv), and an n x n matrix V. If m and n are both large, these matrices can be slow to compute. If only the top few singular values and vectors are computed (i.e. the 'thin' SVD) it would be much faster. Thanks, Sajid From komarek.paul at gmail.com Mon Nov 16 13:18:59 2009 From: komarek.paul at gmail.com (Paul Komarek) Date: Mon, 16 Nov 2009 10:18:59 -0800 Subject: [Research] Thin SVD code in auton codebase? In-Reply-To: <3875.128.2.182.68.1258391437.squirrel@webmail.cs.cmu.edu> References: <3875.128.2.182.68.1258391437.squirrel@webmail.cs.cmu.edu> Message-ID: <459f38470911161018i3e5dc9e8j995c398a1da6b514@mail.gmail.com> I think I checked in a modified fortran version, that I had tried to make callable from C. It seemed like there was some issue, but I don't recall for sure. It was wicked-fast, and supported partial decomposition. Unfortunately, I can't remember the name of it, but the author had "monk" in his name and there might be some f2c or other fortran files in the directory -- maybe play with grep. Paul On Mon, Nov 16, 2009 at 9:10 AM, Sajid Siddiqi wrote: > Hi, > > Does anyone know if there is code for doing a thin SVD (i.e. top k > singular values and vectors) in the auton codebase? > > The svd.c code in /utils seems to return the full SVD component matrices > for an m x n matrix, i.e. a m x n matrix U, a ?n x n diagonal matrix S (as > a dyv), and an n x n matrix V. If m and n are both large, these matrices > can be slow to compute. If only the top few singular values and vectors > are computed (i.e. the 'thin' SVD) it would be much faster. > > Thanks, > Sajid > > _______________________________________________ > Research mailing list > Research at autonlab.org > https://www.autonlab.org/mailman/listinfo/research > From komarek.paul at gmail.com Mon Nov 16 13:20:54 2009 From: komarek.paul at gmail.com (Paul Komarek) Date: Mon, 16 Nov 2009 10:20:54 -0800 Subject: [Research] Thin SVD code in auton codebase? In-Reply-To: <459f38470911161018i3e5dc9e8j995c398a1da6b514@mail.gmail.com> References: <3875.128.2.182.68.1258391437.squirrel@webmail.cs.cmu.edu> <459f38470911161018i3e5dc9e8j995c398a1da6b514@mail.gmail.com> Message-ID: <459f38470911161020o1c8c1b85ncea40ef7e2edb4cb@mail.gmail.com> Aha, here it is (author was munk): the name is propack: http://soi.stanford.edu/~rmunk/PROPACK/ It's still one of the top 10 hits for "partial svd fortran", though maybe there's less competition on that search than in the old days... Paul On Mon, Nov 16, 2009 at 10:18 AM, Paul Komarek wrote: > I think I checked in a modified fortran version, that I had tried to > make callable from C. ?It seemed like there was some issue, but I > don't recall for sure. ?It was wicked-fast, and supported partial > decomposition. ?Unfortunately, I can't remember the name of it, but > the author had "monk" in his name and there might be some f2c or other > fortran files in the directory -- maybe play with grep. > > Paul > > On Mon, Nov 16, 2009 at 9:10 AM, Sajid Siddiqi wrote: >> Hi, >> >> Does anyone know if there is code for doing a thin SVD (i.e. top k >> singular values and vectors) in the auton codebase? >> >> The svd.c code in /utils seems to return the full SVD component matrices >> for an m x n matrix, i.e. a m x n matrix U, a ?n x n diagonal matrix S (as >> a dyv), and an n x n matrix V. If m and n are both large, these matrices >> can be slow to compute. If only the top few singular values and vectors >> are computed (i.e. the 'thin' SVD) it would be much faster. >> >> Thanks, >> Sajid >> >> _______________________________________________ >> Research mailing list >> Research at autonlab.org >> https://www.autonlab.org/mailman/listinfo/research >> > From neill at cs.cmu.edu Mon Nov 16 13:25:56 2009 From: neill at cs.cmu.edu (Daniel B. Neill) Date: Mon, 16 Nov 2009 13:25:56 -0500 (EST) Subject: [Research] one more Lab meeting this week In-Reply-To: <4B010182.5040208@cs.cmu.edu> References: <4B010182.5040208@cs.cmu.edu> Message-ID: Here's a more precise agenda for Wednesday's meeting: Daniel- Introduction to the NSF-funded Fast Subset Scan project (5 min) Skyler- Fast Graph Scan for Scalable Detection of Arbitrary Connected Clusters (practice talk for ISDS conf., 15 min + 5 for questions) Ed- Fast Subset Scan for General Multivariate Datasets (15 min + 5 for questions) Daniel- Fast Subset Sums for Multivariate Bayesian Scan Statistics (practice talk for ISDS conf., 15 min + 5 for questions) Best regards, Daniel On Mon, 16 Nov 2009, Artur Dubrawski wrote: > Dear Autonians, > > On Wednesday, we will be in for a rare treat. > Three tenors, Skyler, Ed, and Daniel, will be performing a piece in > three movements > on the topic of "Fast Subset Scanning and Summing for Scalable Pattern > Detection". > This will be a dress rehearsal for the upcoming tournee in Florida. > Please come in big numbers, and enjoy. > > Time: Noon on Wednesday, Nov 18. > Karen W. will follow up with the location info. > > Artur > > _______________________________________________ > Research mailing list > Research at autonlab.org > https://www.autonlab.org/mailman/listinfo/research > > From awd at cs.cmu.edu Tue Nov 17 09:55:18 2009 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Tue, 17 Nov 2009 09:55:18 -0500 Subject: [Research] Place: Gates 8102 Re: one more Lab meeting this week In-Reply-To: <4B010182.5040208@cs.cmu.edu> References: <4B010182.5040208@cs.cmu.edu> Message-ID: <4B02B956.7050505@cs.cmu.edu> Hi, The meeting tomorrow will be in Gates 8102. Artur Artur Dubrawski wrote: > Dear Autonians, > > On Wednesday, we will be in for a rare treat. > Three tenors, Skyler, Ed, and Daniel, will be performing a piece in > three movements > on the topic of "Fast Subset Scanning and Summing for Scalable Pattern > Detection". > This will be a dress rehearsal for the upcoming tournee in Florida. > Please come in big numbers, and enjoy. > > Time: Noon on Wednesday, Nov 18. > Karen W. will follow up with the location info. > > Artur > > _______________________________________________ > Research mailing list > Research at autonlab.org > https://www.autonlab.org/mailman/listinfo/research > > From awd at cs.cmu.edu Mon Nov 23 11:20:08 2009 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Mon, 23 Nov 2009 11:20:08 -0500 Subject: [Research] Auton Lab meeting this week Message-ID: <4B0AB638.1010300@cs.cmu.edu> This Wednesday we will have a multi-faceted Thanksgiving Special: Daria Sorokina will present "Multivariate Analysis for Predicting Risk of Microbial Contamination of Food" and Robin Sabhnani will come out with two talks: "Discovering Possible Linkages between Food-borne Illness and the Food Supply Using an Interactive Analysis Tool" and "Detection of Disjunctive Anomalous Patterns in Multidimensional Data". Those are rehearsals before the Syndromic Surveillance conference next week. Meeting time: Noon, November 25th 2009. Place: Karen W. will confirm the exact location. Food: yes See yinz there! Artur From awd at cs.cmu.edu Mon Nov 23 13:22:36 2009 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Mon, 23 Nov 2009 13:22:36 -0500 Subject: [Research] Location: NSH 1507 Re: Auton Lab meeting this week In-Reply-To: <4B0AB638.1010300@cs.cmu.edu> References: <4B0AB638.1010300@cs.cmu.edu> Message-ID: <4B0AD2EC.1070806@cs.cmu.edu> Artur Dubrawski wrote: > This Wednesday we will have a multi-faceted Thanksgiving Special: > > Daria Sorokina will present "Multivariate Analysis for Predicting Risk > of Microbial Contamination of Food" > and Robin Sabhnani will come out with two talks: > "Discovering Possible Linkages between Food-borne Illness and the Food > Supply Using an Interactive Analysis Tool" > and "Detection of Disjunctive Anomalous Patterns in Multidimensional Data". > > Those are rehearsals before the Syndromic Surveillance conference next week. > > Meeting time: Noon, November 25th 2009. > Place: Karen W. will confirm the exact location. > Food: yes > > See yinz there! > Artur > > > > _______________________________________________ > Research mailing list > Research at autonlab.org > https://www.autonlab.org/mailman/listinfo/research > > From awd at cs.cmu.edu Sun Nov 29 22:30:49 2009 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Sun, 29 Nov 2009 22:30:49 -0500 Subject: [Research] Special Auton Lab Seminar: Dr. Simon Labov, Tue Dec 01 at 9am Message-ID: <4B133C69.6030500@cs.cmu.edu> (Note: there are two Auton Lab seminars scheduled on Tuesday, the other announcement to follow shortly) *Title* Nuclear Analytics for National Security Simon Labov Lawrence Livermore National Laboratory *Time and Place* Tuesday December 1st 2009, 9:00 AM NSH 1507 *Abstract* Radiation detection systems are currently deployed at ports of entry to scan incoming vehicles and cargo for radiological or nuclear threats, and there is ongoing development work to further improve their effectiveness. Mobile detection systems and radiation sensor networks are also being developed to address threats that do not pass through these ports, are missed by the port systems, or are created within the country. In some cases new detection technology hardware can provide a cost-effective performance improvement for these systems. But in all cases, improved analysis techniques and algorithms can provide significant performance improvements with only minimal cost of deployment. Nuclear analytics goes beyond the standard signal processing techniques used in the conventional analysis of nuclear measurement data and brings more comprehensive analysis approaches to the problem. Machine learning, expert systems, data mining, sensor fusion and data fusion are some of the techniques that can be used to enhance threat sensitivity with the extremely low false alarm tolerance required for nuclear threat detection systems. This talk will introduce the nuclear security problem, review current and emerging detection systems, and discuss on-going work in nuclear analytics for nuclear screening, mobile search and distributed sensor network systems. This work was supported by the Domestic Nuclear Detection Office of the Department of Homeland Security and performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. *Speaker's Bio* Simon Labov is the Associate Program Leader for Detection Systems in the Nuclear Detection and Countermeasures Research Program at Lawrence Livermore National Laboratory (LLNL). He received a B.S. in physics from Stanford University in 1980, and a M.A. and Ph.D. in astronomy from U.C. Berkeley in 1984 and 1988 respectively. Before coming to LLNL, his work focused on X-ray and UV instrumentation for astronomy including filters, optics and detectors. He designed and built a grazing-incidence extreme ultraviolet spectrometer, and flew it on a sounding rocket to measure diffuse emission from the hot gas in the interstellar medium. In 1987 he joined LLNL as a postdoctoral researcher and helped initiate a program to develop high-resolution energy-dispersive X-ray detectors that operate at very low temperatures. As a career physicist at LLNL he built this program to include optical and gamma-ray detectors, and ion detectors for time-of-flight mass spectrometry with large biomolecules. In 1999 he formed the Radiation Detection Center (RDC) to create a focus for detector development work throughout the laboratory. He served as Director of the RDC from 1999 to 2006. From awd at cs.cmu.edu Sun Nov 29 22:31:20 2009 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Sun, 29 Nov 2009 22:31:20 -0500 Subject: [Research] Special Auton Lab Seminar: Dr. Mark Wrobel, Tue Dec 01 at 12noon Message-ID: <4B133C88.6050006@cs.cmu.edu> (Note: there are two Auton Lab seminars scheduled on Tuesday) *Title* Challenges and Opportunities in Research Supporting the Global Nuclear Detection Architecture Mark C Wrobel, PhD, CHP Program Manager Transformational & Applied Research Directorate Domestic Nuclear Detection Office U.S. Department of Homeland Security *Time and Place* Tuesday December 1st 2009, *12:00 noon* NSH 1507 *Abstract* The seminar will discuss the current organization and mission priorities of the Domestic Nuclear Detection Office, current challenges in developing the Global Nuclear Detection Architecture, research priorities necessary to support the Architecture, and provide an overview of DNDO's current research strategy and portfolio. *Speaker's Bio* Dr. Mark Wrobel is a senior health physicist and engineer with twenty-three years of experience in radiation protection, radiation measurements, and radiological engineering. He currently serves as a program manager within the Domestic Nuclear Detection Offices? (DNDO) Transformational and Applied Research Directorate, where he manages research in advanced nuclear detection technologies. Dr Wrobel came to the DNDO from the Defense Threat Reduction Agency (DTRA) where he managed the passive radiation detection technologies branch and a $21M/year strategic research portfolio. Research efforts encompassed new scintillators and semiconductor detection materials, gamma-ray spectroscopy and imaging systems, neutron detection and spectroscopy systems, and detector signal processing and modeling. Prior to DTRA, Mark served 21 years with the U.S. Air Force, retiring with the rank of Lt Col. During his military tenure, Dr Wrobel served as the senior consultant to U.S. Air Force Surgeon for health physics, led management of the AF Master Materials License issued through the Nuclear Regulatory Commission, and was the career field manager for 45 health and medical physicists in the Service. His experience in research management included his role as the health effects officer for the Joint Non-Lethal Weapons Directorate (JNLWD) where he established DoD?s first applied research program in the health effects of non-lethal systems. His other assignments included a number of consultant position in operational and medical health physics, nuclear and radiological consequence management, non-ionizing radiation protection, and environmental remediation. He also served as Chief, Radioanalytical Services and Deputy Chief, Air Force Radiation Assessment Team (AFRAT). LtCol Wrobel?s active duty medals included the Defense Meritorious Service Medal, and three Meritorious Service Medals. Mark received his doctorate in health physics through the University of Michigan, and his B.S. and M.Eng. in nuclear engineering through Rensselaer Polytechnic Institute. He is a member of the Health Physics Society, and has been certified for 20 years by the American Board of Health Physics.